Associations with disease identified by genome-wide association studies (GWAS) must be replicated and refined to validate causative variants. In the Wellcome Trust Case Control Consortium (WTCCC) GWAS using 14 500 non-synonymous single nucleotide polymorphisms (nsSNP), rs11062385 (a nsSNP in JARID1A) showed nominal association with ankylosing spondylitis (AS) (P=0.0006, odds ratio (OR)=1.26, 95% confidence interval (95% CI)=1.1-1.4). To replicate and refine the association of JARID1A, rs11062385 was genotyped in 730 further cases and compared with allele frequencies in non-AS disease cohorts typed by WTCCC. We replicated the initial association (P=0.04, OR=1.16, 95% CI=1.01-1.34) and identified a strengthened association with AS in a meta-analysis of this new study combined with the original WTCCC study (P=0.0001, OR=1.21, 95% CI=1.10-1.33). We also genotyped nine further intronic tagging SNPs in JARID1A in 1604 AS cases and 1020 new control samples, but none was associated with AS. JARID1A or a locus in strong linkage disequilibrium with it is a positional candidate for susceptibility to AS.